Cardiac screening protocol (with screening results). CAD indicates coronary artery disease; AAA, abdominal aortic aneurysm; PTCA, percutaneous transluminal coronary angioplasty; and CAB, coronary artery bypass.
Life-table survival curve after abdominal aortic aneurysm repair.
Won A, Acosta JA, Browner D, Hye RJ. Validation of Selective Cardiac Evaluation Prior to Aortic Aneurysm Repair. Arch Surg. 1998;133(8):833-838. doi:10.1001/archsurg.133.8.833
Copyright 1998 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.1998
To evaluate perioperative and long-term morbidity in patients undergoing selective evaluation of coronary artery disease prior to abdominal aortic aneurysm (AAA) repair.
University and Veterans' Administration medical centers.
One hundred eighty-nine consecutive patients undergoing AAA repair between January 1989 and September 1996 were selectively evaluated for coronary artery disease and assigned to 1 of 3 groups: group 1, no abnormal cardiac history, normal electrocardiogram; group 2, minimal symptoms, history of myocardial infarction (MI), older than 70 years, diabetes mellitus, or congestive heart failure; or group 3, severe or unstable angina, ventricular dysfunction.
Group 1 patients proceeded to AAA repair without further workup. Group 2 patients underwent pharmacologic or exercise stress testing followed by coronary angiography and intervention as required. Group 3 patients went directly to coronary angiography and intervention as needed.
Main Outcome Measures
Perioperative MI, arrhythmias, or death. Long-term follow-up measures included MI and death.
Adequate documentation was available on 171 patients. Twenty-four patients (14%) were in group 1. Of 136 patients (79.5%) in group 2, coronary angiography was performed in 36 (26%), followed by percutaneous transluminal coronary angioplasty (PTCA) in 9 (7%) and coronary artery bypass (CAB) in 5 (4%). Of 11 patients in group 3, 3 (27%) each received PTCA and CAB. Remote CAB or PTCA had been performed in 32 (19%) and 12 (7%) patients, respectively. Two perioperative deaths (1.1%) occurred in the 189 patients, one due to MI in a group 2 patient. There were 2 (1%) nonfatal MIs, both in group 2 patients who had no preoperative intervention. Arrhythmias and/or congestive heart failure occurred in 17 (9%) cases, 7 (39%) having had recent coronary revascularization (P=.001). By univariate analysis, only preoperative renal dysfunction predicted perioperative complications (P=.03) Overall survival by life-table analysis was 87.9% and 69.7% at 3 and 5 years, respectively.
Coronary artery disease is common in patients undergoing AAA repair, with 35.7% having preoperative coronary revascularization at some point. Selective preoperative coronary artery disease screening achieves excellent perioperative and late results in this population.
CORONARY ARTERY disease (CAD) is common among patients with abdominal aortic aneurysm (AAA), with as many as 31% of patients found to have significant disease by angiography.1 Myocardial infarction (MI) remains the leading cause of perioperative and late death in patients undergoing AAA repair. Various studies have documented that cardiac complications (MI, arrhythmia, congestive heart failure) are responsible for 38% to 70% of the operative deaths following AAA surgery.2- 4 Late survival rate is also significantly affected by heart disease among this patient population. Johnston5 determined that the 5-year heart-related mortality rate was 14.3% in patients who had undergone aneurysm repair, as compared with 6.4% for the normal age- and sex-matched Canadian population.
Various approaches have been advocated with the goal of reducing perioperative cardiac mortality and prolonging life in patients with AAA. Early efforts in the 1970s attempted to define clinical risk factors that could be used to predict perioperative mortality; this led to the development of clinical scoring systems such as the Goldman risk index6 and the Detsky index.7 Unfortunately, the use of these systems alone has not proven to be sufficiently accurate in predicting postoperative cardiac events.8 Widespread availability of coronary angiography has led some authors to recommend routine angiography in all AAA patients.9,10 In view of the prohibitive cost of routine angiography, as well as the associated morbidity of the procedure, most surgeons have adopted an approach of selective screening of AAA patients for CAD. Studies based on these screening algorithms have demonstrated perioperative mortality and morbidity rates ranging from 1.2% to 2.0%.11- 13 In comparison, a review of outcome data from centers without specified screening protocols yields somewhat higher cardiac-related mortality rates of 2.6% to 3.6%.2,3,14- 16 This study reports the perioperative and long-term morbidity in patients undergoing selective evaluation for CAD prior to AAA repair.
Between January 1989 and September 1996, 189 patients underwent AAA repair at the University of California Medical Center, San Diego, or San Diego Veterans' Administration Medical Center. Retrospective review was performed on the 167 hospital records available, and telephone interviews were conducted to obtain information on the remaining 22 patients without available records.
In both institutions, patients undergoing AAA repair were stratified for cardiac evaluation by placement into 1 of 3 groups: (1) negative cardiac history (ie, no history of MI, cardiac revascularization, angina, or congestive heart failure) and normal electrocardiogram; (2) history of minimal stable angina and/or CAD risk factors (ie, age >70 years, history or electrocardiographic evidence of previous MI, comorbidities including diabetes mellitus, hypertension, chronic obstructive pulmonary disease, or smoking); or (3) symptomatic CAD (angina refractory to medical therapy, previous MI with sustained ventricular dysfunction).
Patients in group 1 proceeded directly to surgery without any further workup for CAD. Those in group 3 were sent for cardiac catheterization and subsequent revascularization, coronary bypass (CAB), or percutaneous transluminal coronary angioplasty (PTCA), as deemed appropriate, prior to AAA repair. Patients classified into group 2 underwent less invasive studies such as exercise treadmill testing or Persantine-thallium scan, and on the basis of those results went on to coronary angiogram if indicated (Figure 1).
Operations were performed in virtually all cases by general surgical residents under the supervision of an attending vascular surgeon. Pulmonary artery catheters and arterial lines were routinely employed. Epidural catheters, autologous blood donation, and autotransfusion were used liberally.
A computerized database was constructed to include age, sex, medical history (history of hypertension, diabetes mellitus, myocardial infarction, congestive heart failure, chronic obstructive pulmonary disease, and renal dysfunction), surgical history (including history of CAB, PTCA, or carotid endarterectomy), preoperative cardiac workup, prior cardiac interventions, postoperative complications (fatal vs nonfatal MI, arrhythmias, congestive heart failure, pneumonia, prolonged intubation, urinary tract infection, sepsis, or distal thrombosis within 30 days of surgery), and long-term follow-up. The primary endpoints were early and late cardiac morbidity or mortality. Univariate analyses were performed to identify predictors of postoperative complications and death. Life-table analysis was performed to determine long-term survival.
The study population consisted of 189 patients who underwent AAA repair. Data were incomplete on 22 of these patients, and consequently the denominator of some of the fractions is not 189. The mean age for the entire group was 69 years, with a range from 51 to 89 years. There were 11 women (6%) included in the study, who had a comparable mean age of 72.7 years (range, 62-86 years). Thirty-two patients had a history of distant CAB (>6 months prior to AAA surgery) and 12 had undergone previous PTCA; 3 patients had a prior history of both CAB and PTCA. Other common comorbidities included hypertension in 113 patients, diabetes mellitus in 15 patients, renal dysfunction (creatinine level >133 µmol/L [>1.5 mg/dL]) in 40 patients, chronic obstructive pulmonary disease in 44 patients, and smoking (>10 pack-year history) in 120 patients.
Data regarding cardiac evaluation was available in 171 patients. As illustrated in Figure 1, 14% (24/171) of our patients had no cardiac history or electrocardiographic evidence of cardiac disease and underwent surgery without further workup. The majority of patients (79.5% [136/171]) were assigned to group 2, and underwent noninvasive testing. The most commonly performed noninvasive test was a Persantine-thallium scan, while some patients underwent exercise treadmill studies or dobutamine echo at the discretion of the cardiologist. Among these patients, 100 had either normal or insignificant findings and went on to surgery without additional intervention. Thirty-six, however, were believed to have significantly abnormal test results and underwent cardiac catheterization: 9 of these patients had PTCA while 5 underwent CAB prior to AAA repair. The remaining 18 had either normal coronary angiograms, lesions not felt to require treatment, or small-vessel disease not amenable to revascularization. Of the study group, 6.4% (11/171) were believed to be at substantial cardiac risk and went directly to angiography at the outset. Among these, 3 subsequently received PTCA while 3 others had CAB in preparation for their vascular surgery. Thus, a total of 20 patients (12 PTCA and 8 CAB) underwent cardiac intervention based on their preoperative cardiac workup.
In this series of 189 elective AAA repairs, there were 78 patients who experienced postoperative complications (Table 1). Some patients experienced more than 1 complication. Among these complications were 2 nonfatal MIs (1%) and 1 fatal MI (0.5%). Perioperative MI in this series was defined as electrocardiographic changes consistent with ischemia in conjunction with elevated creatine kinase isoenzyme levels. One of the patients with a nonfatal MI had a medical history notable only for a history of smoking with an entirely normal Persantine-thallium test. The other patient had a history significant for exertional chest pain, hypertension, and distant tobacco use, with a fixed defect on preoperative Persantine-thallium. Both were white men younger than 70 years. A single patient suffered a fatal MI following a stormy postoperative course due to a blood transfusion reaction complicated by adult respiratory distress syndrome. This patient had no history of cardiac disease but did have a positive noninvasive test and subsequent catheterization that showed 2-vessel CAD not believed to require revascularization. No MIs occurred in patients who had either remote or immediate preoperative coronary revascularization.
χ2 Analyses were performed to identify preoperative comorbidities associated with postoperative complications. The only factor predictive of complications was a history of renal dysfunction (P=.03); nonischemic cardiac events (arrhythmia and congestive heart failure) were the most common complications for this subset of patients. Preoperative atherosclerotic risk factors of hypertension, diabetes mellitus, smoking history, or prior MI were not associated with statistically significant difference in postoperative morbidity. Interestingly, a history of recent coronary revascularization was found to be predictive of nonischemic cardiac complications (arrhythmia and/or congestive heart failure), with 7 of 17 patients experiencing these complications having had recent revascularization (P=.001).
There were 2 postoperative deaths, yielding a 30-day operative mortality rate of 1.1%. There was a single cardiac death, as described earlier. The remaining early death was due to multisystem organ failure after major hemorrhage secondary to intraoperative iliac vein injury.
Late follow-up was available in 160 patients, of whom 29 had died (Table 2). Cardiovascular disease (MI or stroke) was the most common known cause of late death. Of note, however, the cause for 20.6% of these deaths remains unknown. Life-table analysis showed overall survival at 3 years and 5 years of 87.9% and 69.7%, respectively (Table 3 and Figure 2). There was no statistically significant difference in survival between patients without clinical evidence of CAD, with a history of remote coronary revascularization, or with recent revascularization.
Coronary artery disease is common in patients undergoing AAA repair, with 35.7% of patients in this series having a history of coronary revascularization at some point prior to operation. The 1.1% operative mortality reported here is much improved in comparison with the 4.8% to 9.6% mortality rates of historical studies prior to the advent of pharmacologic stress testing.2,14- 18 Our results are comparable to the 1.2% to 4.8% mortality rates reported in other studies using similar screening protocols (Table 4). Comparison with historical controls can be informative but must also acknowledge that improvements in surgical technique, anesthetic care, and perioperative management have also contributed to the decline in early mortality.
In contrast with other studies showing MI to be a significant cause of postoperative death associated with repair of AAA, fatal MI caused only 1 of 2 early deaths in the present series. Notably, neither group 1 nor group 3 patients suffered a postoperative MI. The group 1 patients cleared immediately were safely and appropriately identified as low risk based solely on clinical criteria. Although few, patients in group 3 at obvious high risk who proceeded directly to catheterization had survival outcomes identical to those with no or clinically insignificant CAD. Among these group 3 patients, 55% (6/11) subsequently underwent preoperative intervention, which suggests that preoperative revascularization may offer a protective effect against MI (although not for arrhythmia or congestive heart failure). Alternatively, these patients may have been more closely monitored perioperatively because of their clinical history. All 3 perioperative MIs (1 fatal, 2 nonfatal) occurred among group 2 patients. Of those patients classified as group 2, 26% underwent coronary angiography and 10% ultimately were determined to require preoperative intervention. Therefore, the clinical criteria used to select patients for noninvasive testing lacked specificity.
The 5-year survival rate (69.7%) reported here is comparable to previous reports in the literature. Feinglass et al22 calculated a 64% 5-year survival rate in a multicenter Veterans Administration study. Johnston et al5 reported a 67.7% survival rate at 5 years, and Starr et al23 a rate of 77%. Of note, the study by Starr et al came from a center where 81% of the patients underwent preoperative coronary angiography.
This study had the usual sources of error inherent in a retrospective review: inconsistencies in interpretation of cardiac evaluation tests, inconsistencies in adherence to the screening protocol, dependence on adequate documentation, and loss of follow-up among patients. These problems were recognized and minimized to the extent possible. An important source of error that merits further discussion was the lack of routine screening for postoperative MI. Because evaluation for MI by electrocardiogram and enzyme studies was performed only in symptomatic patients, silent MIs would not have been detected and subsequently the true incidence of MI was likely underestimated. Additional data would also be desirable in the population of patients with AAA who ultimately did not undergo surgical repair as a result of the cardiac screening protocol. Because this analysis included only those who actually underwent AAA repair, excluded from the study population were those patients who were considered too high risk to undergo surgery, dropped out because of morbidity or mortality from cardiac evaluation or intervention, or were simply lost to follow-up because of the delay. Some patients will clearly fail to have their aneurysm treated because of the process and complications associated with cardiac evaluation. Recently, Mesh et al24 found a CAB morbidity rate of 41% among patients with peripheral vascular disease, considerably higher than the population without peripheral vascular disease. This phenomenon remains to be adequately addressed in the literature,24,25 but is relevant to the issue of the extent of preoperative cardiac evaluation.
In conclusion, the results reported herein support those of other contemporary studies showing a marked reduction in cardiac mortality rates with the development of strategies to address CAD in AAA patients. Unfortunately, our study did not identify clinical factors that would improve the yield of noninvasive testing. As currently employed, selective screening for CAD offers a reasonable approach to achieving excellent early and late results in this population. Indeed, some authors have suggested that cardiac mortalities will no longer continue to be the leading cause of death after aneurysm surgery.26
James J. Peck, MD, Portland, Ore:The authors advocate in this paper a middle-of-the-road approach for preoperative cardiac evaluation of abdominal aortic aneurysm reconstruction. The spectrum ranges from mandatory coronary angiography proposed by Norm Hertzer at The Cleveland Clinic to the belief of Lloyd Taylor, Rich Yaeger, and John Porter at Oregon Health Science University that modern vascular surgery is safe enough to preclude even noninvasive testing in the vast majority of patients. This is due to a number of factors, the most important of which is improved modern anesthetic care, standardized vascular operative techniques, and improved perioperative critical care.
With a well-trained cardiac anesthesiologist and a critical care team, do we need to spend the money preoperatively for this cardiac testing? Nearly 80% of the patients in this study underwent noninvasive testing and 11% had coronary angiography. Thus, as the authors point out in their paper, the clinical criteria lack specificity.
What patients then need to be evaluated? We can definitely exclude certain patients: first, patients with prior coronary artery bypass graft without recurrence of angina; second, functionally active patients with their only clinical cardiac monitor or risk factor being age older than 70 or diabetes alone; third, patients with a favorable noninvasive test within a year before their surgery; and fourth, in patients with symptomatic or very large abdominal aortic aneurysms greater than 7 cm, the delay in cardiac testing would be inappropriate. The fundamental question here is whether or not further knowledge of the patient's coronary anatomy would influence the short-term management of these patients. We are not looking for occult cardiac disease but coronary disease that on its own merit requires mechanical intervention. Who then are the candidates? In my opinion, it is a very few patients who present electively with small or moderately sized aneurysms, less than 7 cm, in which selective evaluation is not so much a vehicle to prevent perioperative cardiac complications but an assessment of the patient's short- and long-term cardiac prognosis; ie, is abdominal aortic aneurysm repair appropriate? These few patients would include patients who need complex aortic reconstruction, particularly those in which supraceliac clamping is anticipated. Secondly, patients with severe pulmonary disease or renal dysfunction who have an increased overall operative risk. And, third, patients with known severely depressed left ventricular function or unstable angina, the group 3 patients in this series.
Abdominal aortic aneurysm repair is by definition prophylactic. It is rational only if patients have a reasonable life expectancy. The goal of risk stratification in contemporary practice is to detail those patients with left main disease and/or triple-vessel coronary artery disease in which coronary revascularization will prolong life. Dr Won and Dr Hye and their colleagues report a remarkable 1.1% mortality with a 0.5% cardiac mortality. Moreover, 98% of these cases were performed by general surgical residents with attending vascular surgeon supervision. This report is as good as it gets in the literature. No one has a better mortality in the literature. The Canadian multicenter trial had more than 660 patients with a reported 4.8% mortality, and two thirds of those deaths were due to cardiac events.
Why, then, are these results so good? The answer is that this is a highly selected group of patients. They represent only the patients who were operated on for abdominal aortic aneurysm reconstruction. Thus my questions: How many abdominal aortic aneurysm patients underwent cardiac screening? What is the denominator here? How many patients were found to be unacceptable operative risks? What was the morbidity and the mortality of the cardiac intervention; ie, how many patients died or were disabled by the cardiac surgery so that they were not now surgical candidates for aortic surgery?
Pulmonary artery catheters and arterial lines were used in all patients. Based on this study, is invasive monitoring necessary and will it improve results in patients with no evidence of coronary artery disease, ie, group 1 patients or patients with normal coronary anatomy?
In conclusion, I strongly concur with the authors that selective cardiac evaluation is a surgeon's responsibility. It is inappropriate to delegate this decision to medical or cardiac consultants. The surgeon has the perspective on the urgency and the relative and absolute indications for abdominal aortic intervention. Moreover, the same selective cardiac evaluation applies to all patients who are going to undergo any major general surgical operations.
F. Don Parsa, MD, Honolulu, Hawaii: I would like to ask the presenters whether or not they denied triple A (AAA) surgery to any patient. My understanding was that everyone was evaluated and that even the sickest of all patients within group 3 underwent coronary bypass or some form of cardiac procedure prior to AAA. Are you turning away any patients because of their medical conditions?
Junaid Khan, MD, San Francisco, Calif: Did you have any problems convincing your cardiac surgeons to operate on asymptomatic patients in group 2? This question comes up quite often at cath conference for patients undergoing hip replacement surgery, for example, and carotid disease more commonly. It is never a clear decision in our cath conferences. I was wondering if there was any difficulty getting the cardiac surgeons to agree to operate on patients who were asymptomatic in group 2.
Dr Hye: Dr Peck, I agree with your assessment for the most part, and I will deal with each of your questions in sequence.
First of all as to your general comments, I would agree that patients who have had a recent CAB and are not symptomatic do not require noninvasive evaluation. Similarly the active asymptomatic patient, which really represents our group 1 patient, does not require extensive evaluation. Additional patients with large aneurysms and those who have symptomatic aneurysms have indications for operation that are urgent enough that one should not delay too much in getting to them. I also agree that the major improvements in morbidity and mortality, particularly cardiac morbidity and mortality, in these patients is a consequence of careful operative technique, excellent intraoperative care, and postoperative critical care.
Whether AAA repair requires a cardiac anesthesiologist I think is debatable. The institution where I currently work does the vast majority of these cases with nurse anesthetists supervised by anesthesiologists. There is a high volume and the results are very comparable with the results that we have had at University of California, San Diego. The other aspect of this issue of course is that while the perioperative mortality might be excellent using this approach, the other value of screening that might be lost is any benefit in prevention of late death; that is, identifying that patient with critical left main disease or 3-vessel disease that requires intervention.
One of the things that concerned me when I looked at these data was that although we achieved excellent perioperative results, our long-term survival was really no better than that of series published 10, 15, and 20 years ago. So clearly our patients are not surviving longer and there still is potential improvement to make in this area. I'm not convinced that more aggressive preoperative screening is the answer, however.
As you pointed out and as we state in the manuscript, one of the weaknesses of the study is that we don't know the denominator. We did not have a mechanism in place at the time that we performed this study to track the patients who had abdominal aortic aneurysms who were sent for cardiac evaluation and were subsequently lost to follow-up, though I believe the number is low. We operate on most patients with aneurysms 5 cm or larger, and the majority of our exclusions are due to extreme age, dementia, coexisting cancer, or other medical conditions. In fact, we exclude very few patients as a consequence of their coronary or cardiac disease.
We also don't know the exact morbidity and mortality of the cardiac interventions in our patients. We have excellent cardiologists and a cardiac surgery group and their results are comparable to those of the best centers in the country, but we in fact don't know the individual results in our patients. The patients who returned to us certainly did well with their cardiac interventions; however, there undoubtedly were some patients who had complications and did not return.
It is pertinent to that issue to recall that there are a number of reports in the literature, most recently from Case Western University this year, showing that the morbidity and mortality of performing coronary bypass in patients with aortic aneurysms is higher than that in a population without aneurysms or other peripheral vascular diseases. In fact, the complication rate in the Case Western series in patients with infrarenal aneurysms undergoing prophylactic coronary bypass was almost 40%, so that is certainly important to keep in mind.
Dr Peck, I think that we did overmonitor these patients. Currently, at Kaiser in San Diego, we use pulmonary artery catheters in fewer than 10% of patients and again the results are quite comparable, so your points are well taken in that regard. We do need to be more selective. As far as the other discussants are concerned, we certainly did have some patients get turned away, and again I think that is the major weakness in the paper. We don't know exactly how many of those patients there were. As far as the question regarding getting the cardiac surgeons to operate on patients who are asymptomatic, we have an aggressive cardiac surgery group and in fact this study was prompted to some extent by their belief that patients should undergo routine coronary angiograms prior to aortic aneurysm repair. That hadn't been our practice, and we don't believe that should be the practice, but we have not had a problem getting them to operate on asymptomatic patients who needed it.
This study adds to the growing literature supporting a selective approach to cardiac evaluation in these patients. We need to further refine the criteria in order to improve the specificity, but how and when that will be accomplished remains to be seen. We also need to examine the issue of long-term survival more carefully.
Presented at the 69th Annual Session of the Pacific Coast Surgical Association, Maui, Hawaii, February 16, 1998.
Corresponding author: Robert J. Hye, MD, Department of Surgery, Southern California Permanente Medical Group, 4647 Zion Ave, San Diego, CA 92120.